FiftyOneTacx Neo – The inside story.
Its with some frustration I am creating this thread for reasons which will become apparently later. The Tacx Neo (1) is likely the best trainer units on the market still today; providing silent operation, fantastic integration with both Bluetooth & ANT protocols, a class leading maximum power resistance, road feel, along with a stable and trustworthy read of actual power figures while requiring zero calibration. It functions with 1 moving part and two small fans which means there are no secondary pulleys, belts, tensioners, adjustments or otherwise to deal with. Even trainers almost 4 years on can only claim a few of these things. I intend to dig in a little here to uncover what makes this unit so capable, along with highlighting some of the short comings in its design. I will then move to part 2 of this teardown where things will get more interesting from a cyclists perspective.
A little about me to begin with. Former mechanical designer with 14 years’ experience in electrical and mechanical pieces. Mining, material handling, bodywork, electronic packaging, radio comms, and the vehicle design. I also think I’m a good cyclist. I don’t claim to know anything or much at all and at any stage in this write up am happy for someone with more experience to tell me I’m wrong about anything I am claiming.
The obligatory warnings-
I DO NOT recommend opening, repairing, modifying, tampering or otherwise attempting to fix or service these units nor units like this. You should consult with the manufacturer as to the best course of action which is likely to be in consultation with your nearest approved TACX dealer or distributer. The risks you face by messing with this are large electrical shocks, possible electrocution, broken bikes, broken you, fire, an unrepairable unit, death or worse than that telling your wife you ‘fixed’ the $2000 trainer so well it needs to be sent away for professional repair. In short it is unwise to disregard the user manual (the manufacturers opinion) and attempt anything below. DO NOT open the unit.
However, I am a special kind of stupid and as will become apparent in coming posts there is a reason I am undertaking this.
You’ll require the dedicated flywheel puller to get the wheel out prior to disassembly. YOU MUST use a puller, particularly when you are reinstalling the wheel. The magnets on the Neo are incredibly strong and will snap down on your fingers or get jammed on the shaft potentially damaging the magnets, the shaft, the bearings or other components on its way in. Again, don’t recommend you do this however you will need a flywheel tool. A quick note about the flywheel and the puller; Initially Tacx did not offer the puller to the public, truly making the unit unserviceable. With reports of the infamous grinding noise coming up online, a decision was made to make the tool available to the public and a change to online documentation was made to educate users of the correct way to maintain the unit. The issue is caused in large part from the highly magnetic flywheel and its proximity to the bikes drivetrain. The reality is fully sealing the flywheel with a gasket or similar would be difficult but due consideration wasn’t really given to the full view of how this product will be used. Drivetrains deteriorate, debris can rattle out of calliper brakes, a trainer is picked up to be transported in the back of cars, they are used in sheds or outdoors. Eventually something metallic will make its way towards the magnetic surfaces. The good news is that rectifying any issues is generally easy.
The first thing that becomes apparent when disassembling the Neo is the large number of star bit screws otherwise known as Torx screws securing everything. These are not security Torx screws as denoted with the central pin which is designed to slow down entry to the unit. The screws have quite a rough course thread on them which is designed to bite in to the plastic. The first questions are why so many screws and the type selected, in short there aren’t user serviceable parts inside and with some of the power electronics we will see later, you probably don’t want to stick your sausage in that. As this is a ‘sealed for life’ unit, the self-threading screws bite in to the plastic in first install meaning that each subsequent removal will begin to destroy the plastic too. This isn’t unusual on most consumer products in the modern age, however with the repair movement growing many people don’t expect to be attempting to repair the screw holes. Beginning to back out all the fasteners, it became apparent that they are not all identical with three major sizes used throughout and a couple of one off’s around the place. I found it best in this case to scribe a ‘1’, ‘2’ or ‘3’ next to the hole. ONCE AGAIN YOU SHOULD NOT DISASSEMBLE THESE UNITS, however if you want to give it a go, you’ll want to somehow record what goes where. First step is taking the flywheel off, followed by the screws on the right leg. The two halves of the right leg will separate revealing some plasma cut and pressed channel galvanised pieces. The leg locking mechanisms are simple with commonly available pin stock and plastic washers holding the assembly in place. Anyone with a Neo will know the all too familiar click in to place. The legs are incredibly stiff and offer no major flex while in operation. When the unit is in storage or on the move however, owners will tell you how much of a pain they can be. The legs are also clumsy to unfold ensuring they are all deployed correctly needing to tip the unit around. The two pins can be punched out and the pressings will come free. The leg pressings are quite thin in comparison to the chassis and while a slight bend in the leg won’t effect operation I was surprised to see how easily they were effected by being removed.
The right leg cover is required to come off before the right-side body cover can be removed so repeat the aforementioned ill-advised steps as above. I was lucky enough to have a torx to hex bit I could run all the screwed out with which saves a lot of time, however install requires more caution so keep that in mind.
Generally, you would stop at this point which is WAY TOO FAR already. Under no circumstances should you touch any of the exposed electricals at this point. There is a chance some many contain a large amount of residual charge which could kill you. Not only could it kill you, it will really really hurt the entire time you’re dying so don’t do it.
Main things that can be seen at this stage are the ceramic load bank and rear fan at the back of the unit, the stator & windings, the PCB control board and the front fan all mounted to the chassis.
The stator is secured to the chassis with six machined dowels. These are really the only semi standard fastener heads on the whole unit which many lead people to think they can gain easy access by removing them. DO NOT DO THIS. To remove the back cover, you once again need to remove the covers from the left leg and any other fasteners on that side prior to the tapered cap heads being removed. It would be sage advice to also drive out the leg pins at this point so the unit can sit flat on a bench. The lower body cover doesn’t need to be removed but I’ve taken it off anyway for the sake of a full teardown.
First comment is on the overalls of the internals. Package space has been used reasonably effectively for both component placement and airflow. The chassis is high quality laser cut, folded and welded assembly made from about 5 parts with adequate quality pressing, jigging, welding and finally what appears to be a nickel coat. A reasonable degree of thought was given to how the parts would fit together for manufacture with the front & rear leg mounts repeated in the assembly and consideration to how the stator would mount to the chassis given sufficient thought. The inside of the stator mount is a high-quality machined piece. The windings appear to have a thermocouple embedded in to them to monitor for any overheating, a theme which will be repeated several times through the teardown. Many people have seen the insides of the flywheel before, so I won’t focus on it too much however the bearings are of reasonable size and are heavily sealed. A specialist press and punches would be required to replace them should the need arise. Now that you’ve covered the magnets in swarf and other metallic debris now is a good time to remind you that you really shouldn’t be poking around inside the units and that the best way to keep the flywheel clean is by placing it inside a bag to prevent anything from sticking to it.
Moving to the load bank, another thermocouple can be seen riveted to the top along with a thematic one-time fuse connecting both sides of the bank. I am unsure why the two halves have been joined in this way, however I imagine once the unit is in operation it would be able to determine if the fuse has gone with use of a resistance calculation and the overheating of the remaining bank. Again, Neo owners can tell you all about the sweet smell of success after dropping a major wattage bomb to KOM, with many people looking around the room in ensure they haven’t set fire to anything on the ceramic elements.
The wires all disappear in to the nether regions of the chassis which unfortunately has some unbroken edges. There is probably a low probability of snags due to the nature of how the unit works however getting wires in and out of that space is not optimal. There is also going to be reduced airflow in that lower area which may become apparent later.
Before looking at the brains of the unit, we should look at the bodywork as there is a story to tell. Two stories to tell in fact. The exterior surfaces on this unit are of an A class fine stipple finish of injected but I don’t believe reinforced ABS. This is where the tail of two units comes in to play, the A class is sound in most places with no visible sink marks or blemishes. The inside surfaces are not great on the main body casings however with highly visible tooling marks everywhere. It also shows evidence of reactive design where notches were placed in the inside tools to accommodate some edges and corners of the internal structure. Three relief area’s can also be seen where a modification to the tool was required for either a changed design or a clearance which was not accounted for in final CAD design. Some of this started to make me think that there were two teams working in semi isolation during the development of the final spec unit. My reasoning is that the chassis, electronics are covered by the main body which has a poor-quality internal finish and forced changes. The legs however are a very different story. The leg covers exhibit the same high-quality A-class finish free from blemishes or sink marks, internally however the finish is quite high with only minor injection marks visible. This is a marked difference with virtually no machine marks to speak of. The pressed metal leg internals, rudimentary hinge and locks along with an undisguisable injection part suggest to me that everything below the chassis was farmed out to a contract house to work on the appearance of the product while Tacx set to work on the functional aspects of the unit. There are more clues to this that can be witnessed such as the subtle change which can bee seen between the 2016 and 2017 models allowing for clearance to a flat mount disc brake calliper. There is also no grab or carry handle on the unit making for a difficult time moving the unit and no retention of the legs while in the stored position. I speculate that this indicates a brief was prepared and an external group worked on the physical elements while Tacx focused on the singing and dancing parts (a job Tacx have proven to have done an exceptional job with)
The brains of the operation are the last remaining item with the main control board located to the front. The Neo is a reasonably innovative product in its field with the ability to function with or without externally supplied power, in this case 40 Watts of 48 Volts which is a slightly unusual voltage for most consumer devices. The astute of the masochists that are still reading to this point will note most of the owners of this device will produce well over 5 times that power at a sustained level and therefore Sir Newton’s Third law would suggest the equal and opposite of the rider’s power is not due to the external supply. In fact, we know this as the unit is able to generate a large amount of simulated resistance while unpowered and the external supply further increases that to eye watering levels, enables the downhill freewheeling and road feel. The general packaging of the board is well thought out and space is used effectively with some room remaining for expansion, likely as seen on the Neo2 and its secondary cadence sensor. One thing that has been highlighted by some has been the incredibly thin wires on the unit. For higher voltages a lower cross section of wire can be used as less current is flowing as given by the power law, P=R.I^2 . However, the wires are somewhat small and delicate for the application. The unusual voltages continue with a 3.8v capacitor placed close to the DC in, a 25v cap south of that and 3 100v large caps at the bottom of the board. THIS IS YOUR ANNUAL REMINDER NOT TO MESS WITH THIS! 100v of anything is large, really large. And those are big caps, if they discharge in to you, you’re going to have a bad day. Generally, consumer electronics run on 5v logic and 12v inputs however those voltages don’t seem evident here, instead the 3.8v cap suggests that something else is used here.
Quickly reviewing the rest of the board, the three large caps and the 3 banks of double MOSFETs on the cooling extrusion suggest that this unit is three phase in design allowing for frequency control of the flywheel. A 7th MOSFET on the end could be what enables a timed misfire for the road feel. MOSFETs or “metal–oxide–semiconductor field-effect transistors” are basically electronic gates which allow a larger voltage to flow through to a driven source. Valve amplifiers are now a thing of the past due to MOSFETs and marketing hype can be seen on many consumer goods such as this car stereo on the wall in my shed.
A brief explainer about how the neo likely works. The unit will detect the movement of the flywheel via the rising level of power and begin to synchronise with it, allowing a slight lag while you build up momentum. The unit will be able to monitor the resultant frequency, the voltage and current peaks and through a bunch of feedback calculations slow or lag its frequency. The shifted frequency will mean as the magnet spins past the energised pole, you will be effectively squeezing the magnetism down giving the feeling of constant resistive force. Think about two opposing magnets on a bench. If you push one at a certain speed, the computer in this case is running away at the same speed. However to create the resistance it pushes back at you, bringing the two virtual magnets closer together giving the feeling you’re always fighting to push the two magnets together. And it does this with your own energy. It’s a fight you can’t win.
The board has 4 distinct areas; (1) The input power side, (2) comms, (3) control and finally (4) the power and frequency control. Something that became quite apparent was the overall lack of shielding for the looms and the board itself. Although the flywheel rpm will range between 200 to 400 for most use cases, the 30 poles of the stator mean there will be a lot of noise in the 3-6kHz range with all the magnets and electrical pixies whizzing around inside the unit. On interrogating some of the SMD components, I found evidence that some are ESD and EMI rated, some for over 200 volts of discharge. Although the use case of some of these chips are not generally for low power 3 phase frequency control, they seem to function fine (most of the time) in the noisy environment. I’m guessing the lack of shielding is due to the lower power and voltages required to do the job, and hence lower currents which cause issue. The lower figures will ultimately mean a lower EMI noise emitted meaning your pacemaker should be ok unless some drops a Neo on your chest. The main IC’s are protected with small resistive values or microcaps as required which is normal of most electronics. My experience with shielding requirements was more around mission critical applications with ever more ridiculous voltage standards and completely OTT spec OCD level requirements for containment (which weren’t any good anyway), but I am interested to know how much effort was actually put in to protection. The FCC testing generally checks the board for stray emissions of any chips that are designed either primarily or otherwise for emitting radio signals. Its not so much to see how much your product will be affected.
As the Neo can generate excess power, we know it stores and uses some of it to self-start. There is a drain-back from one of the large caps to a regulator which supplies voltage to the 3.8v cap allowing the system to boot. This is where some of the genius starts to come in. The 3v value is the same as a CR2032 cell which most cyclists would know powers many of the items like your heart strap, power meters, wireless shifters and head units. The Neo’s two main chips are the (2) nRF51422 bluetooth/ANT chip from Nordic and the (3) TMS320F2802x Piccolo 32bit micro controller from Texas Instruments. These two chips have been a fantastic selection and with the effort put in to both firmware, functional programming and development of the suite of smartphone apps make for a good user experience. This really is what makes the Neo as good as it is. Interestingly the chips also seem to have on board temperature sensors, likely for self-diagnostic uses primarily, but as we’ve seen in other areas there are a few other monitoring points, so I do wonder if they are sudo-monitoring the board and adjusting for it. It is also super impressive how low the power consumption is for both components. As for a breakdown of the programming side, we won’t be going there. The suffice to say secret sauce in there works well and generally no issues are experienced getting the Neo to play nice with anything you want to hook up to it.
Focusing in on the comms side, there is a sticker with QR code here indicating the units identification number. It covers the antenna which is screen printed on the PCB. I always marvel at these small antennas due to how functional they are.
Finally, on the board we can see most of the I/O ports with a few more MOSFETs above them, likely control of the cage fans. The three external indicator lights are seen to the top.